Operational Modal Analysis from multi-setup measurements with roving references and Stochastic Subspace Identification

Accurate characterization of the modal parameters of existing structures from ambient vibration data within the field of operational modal analysis requires, in general, the collection of synchronous measurements at multiple points of the structure. However, when an insufficient number of sensors is available, the experimental campaign must be organized by dividing the measurements into subsets. In these situations, approaches like the reference-based Stochastic Subspace Identification method allows for full structural output-only identification with fixed reference sensors. Nevertheless, having common reference sensors for the whole campaign is not always feasible. Thus, this paper proposes a method to merge measurements from different ambient vibration data collection times and locations without a common reference sensor across all sets. To do so, the approach relies on roving references that are common only between successive groups of simultaneously measured positions. The algorithm’s operation is tested and analysed for two different case studies: a first one that uses synthetic data from a numerical model for the dynamic response of a simply supported beam subjected to ambient vibrations, and a second case that employs experimental measurements at an elevated pedestrian crossing. The results show that, in general, the proposed method is able to identify natural frequencies and mode shapes from ambient vibration measurements even with significant noise levels. Thus, this method can serve as a viable and straightforward alternative in situations where maintaining common reference sensor locations through the whole measurement campaign is not possible.